On Sat, Aug 15 2020, Teemu Likonen wrote:

> The following Unicode's bidirectional control chars are modal so that
> they push a new bidirectional rendering mode to a stack:
>
>     U+202A LEFT-TO-RIGHT EMBEDDING
>     U+202B RIGHT-TO-LEFT EMBEDDING
>     U+202D LEFT-TO-RIGHT OVERRIDE
>     U+202E RIGHT-TO-LEFT OVERRIDE

Good stuff -- implementation looks like port of the php code in 

   https://www.iamcal.com/understanding-bidirectional-text

to emacs lisp... anyway nice implementation took be a bit of
time for me to understand it...

thoughts

- is it slow to execute it always, pure lisp implementation;
  (string-match "[\u202a-\u202e]") could be done before that.
  (if it were executed often could loop with `looking-at`
   (and then moving point based on match-end) be faster...

- *but* adding U+202C's in `notmuch-sanitize` is doing it too early, as
  some functions truncate the strings afterwards if those are too long
  (e.g. `notmuch-search-insert-authors`) so those get lost.. 

- what about https://en.wikipedia.org/wiki/Bidirectional_text#Isolates
  (was documented more in some page, cannot find it anymore...)

(what I noticed when looking `notmuch-search-insert-authors` that it uses
 `length` to check the length of a string -- but that also counts these bidi
 mode changing "characters" (as one char). `string-width` would be better
 there -- and probably in many other places.)

(I tried quite a few things, something that could "reset" the stack with
 e.g. one invisible tab, but no go (or that was filtered as I added it
 to `notmuch-sanitize` ;), As a final step I did

  (defun notmuch-sanitize (str)
  ...
  -  (replace-regexp-in-string "[[:cntrl:]\x7f\u2028\u2029]+" " " str))
  +  (replace-regexp-in-string
  +   "[\u202A-\u202E\u2066-\u2069]" ""
  +   (replace-regexp-in-string "[[:cntrl:]\x7f\u2028\u2029]+" " " str)))

just to test-drop those chars. probably not good enough ;/)


Tomi

>
> Every mode must be terminated with with character U+202C POP
> DIRECTIONAL FORMATTING which pops the mode from the stack. The stack
> is per paragraph. A new text paragraph resets the rendering mode
> changed by these control characters.
>
> This change adds a new function "notmuch-balance-bidi-ctrl-chars"
> which reads its STRING argument and ensures that all push
> characters (U+202A, U+202B, U+202D, U+202E) have a pop character
> pair (U+202C). The function may add more U+202C characters at the end
> of the returned string, or it may remove some U+202C characters. The
> returned string is safe in the sense that it won't change the
> surrounding bidirectional rendering mode. This function should be used
> when sanitizing arbitrary input.
> ---
>  emacs/notmuch-lib.el | 54 ++++++++++++++++++++++++++++++++++++++++++++
>  1 file changed, 54 insertions(+)
>
> diff --git a/emacs/notmuch-lib.el b/emacs/notmuch-lib.el
> index 118faf1e..e6252c6c 100644
> --- a/emacs/notmuch-lib.el
> +++ b/emacs/notmuch-lib.el
> @@ -469,6 +469,60 @@ be displayed."
>       "[No Subject]"
>        subject)))
>  
> +
> +(defun notmuch-balance-bidi-ctrl-chars (string)
> +  "Balance bidirectional control chars in STRING.
> +
> +The following Unicode's bidirectional control chars are modal so
> +that they push a new bidirectional rendering mode to a stack:
> +U+202A LEFT-TO-RIGHT EMBEDDING, U+202B RIGHT-TO-LEFT EMBEDDING,
> +U+202D LEFT-TO-RIGHT OVERRIDE and U+202E RIGHT-TO-LEFT OVERRIDE.
> +Every mode must be terminated with with character U+202C POP
> +DIRECTIONAL FORMATTING which pops the mode from the stack. The
> +stack is per paragraph. A new text paragraph resets the rendering
> +mode changed by these control characters.
> +
> +This function reads the STRING argument and ensures that all push
> +characters (U+202A, U+202B, U+202D, U+202E) have a pop character
> +pair (U+202C). The function may add more U+202C characters at the
> +end of the returned string, or it may remove some U+202C
> +characters. The returned string is safe in the sense that it
> +won't change the surrounding bidirectional rendering mode. This
> +function should be used when sanitizing arbitrary input."
> +
> +  (let ((new-string nil)
> +     (stack-count 0))
> +
> +    (cl-flet ((push-char-p (c)
> +             ;; U+202A LEFT-TO-RIGHT EMBEDDING
> +             ;; U+202B RIGHT-TO-LEFT EMBEDDING
> +             ;; U+202D LEFT-TO-RIGHT OVERRIDE
> +             ;; U+202E RIGHT-TO-LEFT OVERRIDE
> +             (cl-find c '(?\u202a ?\u202b ?\u202d ?\u202e)))
> +           (pop-char-p (c)
> +             ;; U+202C POP DIRECTIONAL FORMATTING
> +             (eql c ?\u202c)))
> +
> +      (cl-loop for char across string
> +            do (cond ((push-char-p char)
> +                      (cl-incf stack-count)
> +                      (push char new-string))
> +                     ((and (pop-char-p char)
> +                           (cl-plusp stack-count))
> +                      (cl-decf stack-count)
> +                      (push char new-string))
> +                     ((and (pop-char-p char)
> +                           (not (cl-plusp stack-count)))
> +                      ;; The stack is empty. Ignore this pop character.
> +                      )
> +                     (t (push char new-string)))))
> +
> +    ;; Add possible missing pop characters.
> +    (cl-loop repeat stack-count
> +          do (push ?\x202c new-string))
> +
> +    (seq-into (nreverse new-string) 'string)))
> +
>  (defun notmuch-sanitize (str)
>    "Sanitize control character in STR.
>  
> -- 
> 2.20.1
_______________________________________________
notmuch mailing list -- notmuch@notmuchmail.org
To unsubscribe send an email to notmuch-le...@notmuchmail.org

Reply via email to